Keep a conversation going on a dating app: why chats stall
Why dating app chats stall, what conversation research says about reply speed and follow-up questions, and restart messages for a chat that went quiet.
A dating chat usually stalls because the last message gave the other person nothing to pick up. The fix is to put one small detail and one question in each message, to answer the question you were asked before asking a new one, and, when a chat has gone quiet, to restart it by returning to something already said instead of changing the subject. This page explains the two bits of conversation research behind that advice, with the caveats, and then gives restart messages for the situations where chats most often die.
How to keep a conversation going, online and in person
How fast people respond. Emma Templeton and colleagues at Dartmouth recorded 322 spoken, ten-minute conversations among 66 undergraduates in same-gender pairs, most of whom had never met. The typical gap between one person finishing and the next starting was about 200 milliseconds, and conversations with faster response times felt more connected, both for people who had just met and, in a follow-up study, for close friends. The speed of the partner's reply counted for more than the speed of one's own. When the team sped up or slowed down the gaps in recordings of real conversations, listeners who had no part in them judged the sped-up versions as more connected.
That study does not tell you how many minutes to wait before answering a text. The gaps it measured are a fraction of a second in speech, too short for anyone to plan, and the authors flag that their conversations were polite get-to-know-you talk between undergraduates. What does carry over is the finding about the partner: people felt connected when the other person responded to them. In a text chat, that shows up as a reply that visibly uses what you just said. Nobody has measured the delay that works best in a dating chat, and our page on double texting covers the one study of timing around a first date.
Whether people follow up. Karen Huang and colleagues at Harvard had pairs type to each other for 15 minutes in an online chat. People who asked more questions were better liked, and the effect came mostly from follow-up questions, the ones that pick up what the other person has just said. Questions that switched to an unrelated topic went with lower ratings of responsiveness, not higher. In a separate set of 110 speed daters on 4-minute dates, people who asked more follow-ups got more agreements to a second date. The details of how that works as a first message are on our page about first messages that get a reply; what matters here is the topic-switch result, because the usual response to a dying chat is to throw in a new topic.
A third finding, the liking gap described on our first date nerves page, explains why a quiet chat feels worse than it is: Erica Boothby and colleagues found that people systematically underestimate how much a conversation partner liked them. A short reply is more likely a busy evening than a verdict.
Face to face, the first two findings apply more directly, because both were measured in live conversation: Templeton's in speech, and Huang's second-date result on face-to-face speed dates. In person you cannot scroll back, so the follow-up has to come from listening. Our rule for a lull on a date: go back to the last thing they said that had a detail in it, and ask how or why.
Why chats stall, and what to send
| What the chat looks like | What is probably going on | What to send |
|---|---|---|
| They answer in a few words and do not ask anything back | The question gave them nothing to work with | Give something, then ask something narrower |
| You are asking all the questions | It has turned into an interview | Answer one yourself, then invite theirs |
| Both of you reply, but nothing builds | Everything stays on the surface | Return to the last real detail and ask how or why |
| It went quiet for a day or more | Life, or no reason to answer | One light message that refers to something shared |
| The topic ran out | Nothing wrong | Back to an earlier thread, not a new one |
Restart messages
These are suggestions, not tested lines. Replace the details with yours.
After a one-word answer. They: "Pretty good, just tired." You: "Tired from what, work or one of those weeks? I'm on day four of a cold I refuse to acknowledge." It offers a detail about you and asks something they can answer with a word or a story.
When you notice you are interviewing. "I realize I've been firing questions at you. Your turn: ask me anything and I'll answer honestly." It resets the balance and shows you are paying attention to the conversation, not just running it.
When they gave a small detail and you want to build on it. They: "I do some climbing." You: "Indoor or outdoor? And what got you into it?" Two follow-ups, both about their own sentence.
After the topic ran out. "Okay, we've done the weather. Going back to something you said earlier: you were about to tell me about the trip to Lisbon." This is a follow-up on something already in the chat, which is what Huang's results favor over a fresh topic.
After a day or two of silence. "Hi, hope your week's going well. I walked past the dumpling place we talked about and thought of you. Did you ever try it?" Liu and colleagues found across preregistered experiments that people underestimate how much others appreciate being reached out to, particularly when it is a bit of a surprise. Their participants were people in each other's circle, not dating matches, so take it as a reason not to assume your message will be unwelcome, not as proof.
If you have sent one restart and nothing comes back, stop. You cannot make someone answer, and a third message mostly shows how much you wanted one. If you want to end it gracefully rather than let it fade, see saying no without ghosting.
Questions to ask on a dating app
Huang's team sorted the questions in their chats into six types: follow-up, full switch (a new, unrelated topic), partial switch, mirror (asking back what you were just asked), introductory and rhetorical. Follow-up questions were the ones that went with being liked. A list of questions you could ask anyone is, by that definition, a list of switches. So treat the questions below, all ours, as shapes to fill from what the other person has written or said.
From their profile, as an opener. "Your last photo is on a ferry. Where was it going?" or "You're learning Portuguese: for a trip, or just because?"
Once they have told you something. "Wait, you rebuilt the bike yourself? How long did that take?" or "You said you mostly work nights. What does a good night shift look like?"
When it could turn into a date. "You keep mentioning that bakery. Want to go on Saturday morning?" or "Is there a walk near you that you'd recommend? I'd go if you came along."
In our view, the questions that help least are "How was your day?" with nothing attached and anything you could have asked without reading their profile. A mirror question such as "You?" keeps things polite, but on its own it adds nothing. For longer lists by stage, see getting to know you questions and flirty questions.
When the chat is going well
A chat that is going well does not need more topics, it needs a plan. At some point the useful move is to suggest a call or a meeting; the page on how to ask someone out covers timing and wording. For lists of topics and starters, conversation starters has them by situation.
Frequently asked questions
What do you say when a conversation dies?
How quickly should you reply?
Is double texting a problem?
Why do dating app chats fizzle out?
What topics work best in online dating chats?
Sources
- Templeton, Chang, Reynolds, Cone LeBeaumont and Wheatley (2022), Fast response times signal social connection in conversation, PNAS Full text read October 11, 2026. Supports 66 Dartmouth undergraduates in same-gender pairs in 322 ten-minute spoken conversations, a typical response gap of about 200 milliseconds, faster responses going with feeling more connected between people who had just met and between friends, the effect running mainly through how fast the partner responded rather than how fast the person themselves did, the manipulation of response times in recordings, and the authors' own limits: undergraduate participants, polite get-to-know-you talk, a ten-minute horizon. The measure is in milliseconds in speech, so it does not apply to the minutes or hours between text messages. Rechecked October 16, 2026 in the PMC full text (66 participants, 322 conversations, modal response time around 200 ms).
- Huang, Yeomans, Brooks, Minson and Gino (2017), It doesn't hurt to ask: question-asking increases liking, Journal of Personality and Social Psychology Full text read October 11, 2026. Supports the studies in which people typed to each other for 15 minutes in an online chat, the finding that people who asked more follow-up questions were better liked, that follow-up questions went with higher rated responsiveness and full-switch questions (about a new topic) with lower, that follow-up questions alone predicted partner liking, and the speed-dating study of 110 people on 4-minute dates in which more follow-up questions went with more agreements to a second date. Reread October 16, 2026: also supports the six question types the researchers coded (follow-up, full switch, partial switch, mirror, rhetorical, introductory), with follow-up questions defined as following up on a topic the partner had mentioned, almost always in the previous turn, and the speed dates being face to face.
- Boothby, Cooney, Sandstrom and Clark (2018), The liking gap in conversations, Psychological Science (University of Essex repository copy) Abstract read on the repository page October 11, 2026. Supports the finding that after conversations people systematically underestimate how much their partner liked them and enjoyed their company, in conversations of varying lengths.
- Liu, Rim, Min and Min (2023), The surprise of reaching out: appreciated more than we think, Journal of Personality and Social Psychology (Europe PMC record) Abstract read on October 11, 2026. Supports preregistered experiments showing people underestimate how much others appreciate being reached out to, more so when the contact is more surprising and the two people are less close. The experiments involved people in each other's social circle, not dating chats.